Contained herein is material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction of the patent disclosure by any person as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all rights to the copyright whatsoever.
1. Field of the Invention
The present invention is related to the field of computer networking. Specifically, the present invention is related to a method and apparatus for monitoring a dedicated communications medium in a switched data networking environment.
2. Background of the Invention
With reference to FIG. 1, prior art data networks generally utilized one or more shared media hubs, e.g., hub 100. Multiple end user workstations, e.g., workstations 1, 2 and 3, were coupled to a shared communications medium (mediums 10, 20 and 30, respectively) that was, in turn, coupled to a port on the shared media hub. The hub 100 had multiple ports (e.g., ports 11, 21, 31), each coupled to a different shared communications medium. High end workstations stations, or servers, such as file servers or print servers, were also coupled via a dedicated or shared communications medium to a port on the shared media hub.
As the applications running on these data networks became more mission critical, and bandwidth utilization of the shared communications media increased, it became advantageous to monitor, for example, the performance and the error rates of data traffic on the shared communications media to facilitate proper operation of the data network. To accomplish this monitoring, network monitoring devices were configured into shared media hubs, or coupled to the port (41) of a shared media hub via a communications medium (40) as stand-alone devices (e.g., probe 4). In either configuration, the monitoring devices were typically referred to as probes. The probes would promiscuously monitor the data traffic on all shared communications media in the network and look at, for example, performance and error statistics, data traffic patterns and typical data flows across the shared communications media.
As shown in FIG. 2, as performance requirements of prior art data networks continued to increase, and additional performance intensive applications were employed, the shared communications media coupled to the shared media hubs were typically divided into multiple network segments (e.g., network segments 201, 202 and 203) to reduce data traffic on each segment, although all network segments were still in the same collision domain, i.e., the network segments were not electrically isolated. Data communication between these segments generally utilized well known backbone, rather than switching, technology.
As performance requirements continued to increase to meet traffic demands, switches such as switch 220 illustrated in FIG. 2 were used to segment the network into multiple collision domains. Segmenting the network into multiple collision domains so that a data packet from one segment (e.g., segment 201) did not traverse the network to another segment (e.g., segment 202) unless the data packet was destined to a particular device on another segment as determined by, for example, a destination address specified in the data packet.
The problem, however, in monitoring network performance in such an environment utilizing probes was that a single probe was required for each segment in order to promiscuously monitor the data traffic on that segment. With reference to FIG. 3, as the data networks became highly segmented, it became evident that it was impractical to attach a probe to each segment in the network to promiscuously monitor all traffic. Rather, network administrators tended to concentrate probing activities to highly concentrated server farms or segments in the network where the traffic was the busiest, for example, a segment from a switch to a file server. These file servers were typically coupled via a dedicated communications medium to a port on a switch to provide, for example, a data communications rate of 10 megabits or 100 megabits per second, or even 1000 megabits per second, to the file server. Connecting the file server using a dedicated communications medium to the switch 220 formed a single station network segment. In a single station network segment, it was impossible to attach a probe to that segment to promiscuously monitor all network traffic because only a single port was necessarily available to which the segment is coupled to the switch. To overcome this limitation, a multiport repeater was inserted between the switch and the file server, e.g., repeater 233 between workstation 3 and switch 220 in FIG. 3, thereby providing additional ports (on the inserted multiport repeater) to facilitate connection of a probe (e.g., probe 235) into the segment.
Although switch 220 in FIG. 3 shows only six ports for purposes of illustration, it is understood that a switch may have sufficient ports to support, for example, ten or twenty servers. In such a situation, it becomes impractical to attach a repeater between every server and switch port to promiscuously monitor data traffic, due to the increased cost, space, and asset management responsibilities encountered as a result of the additional equipment. In addition, for each communications medium that was to be monitored, that network segment would have to be taken down, the server disconnected from the switch, the repeater inserted into the communications medium, and the server communication reestablished. This process would be highly disrupting to data communications in the network. Moreover, in attempting to diagnose a performance problem, one would be required to shut down the network segment, insert a repeater, and couple a probe to the repeater in order to collect monitoring data. By the time the probe was operable, the performance problem may well have disappeared.
A method and apparatus for monitoring performance in a switched network environment is disclosed. An embodiment of the present invention is coupled between a switch and a network device, for example, a file server, to provide a promiscuous tap into the communications medium between the switch and the network device. The apparatus promiscuously monitors all packets between the switch and the network device without the need for a separate repeater, yielding significant economic savings and eliminating the downtime as may occur in the prior art when performing troubleshooting operations.
A bypass circuit associated with the probe allows promiscuous monitoring of all traffic between the switch and the network device in either direction, and in full duplex mode. Additionally, the bypass circuit eliminates the requirement for a separate repeater between the switch and the network device. Moreover, the bypass circuit is fault tolerant, i.e., if power is disrupted to the probe or if the probe malfunctions, the bypass circuit allows traffic to continue to pass between the switch and the network devicexe2x80x94data traffic is not disrupted in any manner as a result of failure or malfunction of the monitoring aspect of the probe.